Fetch the financial records nobody else covers.
Show Tabular how to pull a statement once. It writes the workflow, then replays it whenever you ask — into a spreadsheet you own, with no aggregator in the middle and no credentials handed over.
- Runs in your browser, in your own session
- Rows go from the page to your sheet
- Your workflows and your rows stay in your Drive
Billing statements
Page 1 of 3| Date | Description | Amount |
|---|---|---|
| 2026-07-01 | Tuition — July | $1,240.00 |
| 2026-06-03 | Late pickup fee | $35.00 |
| 2026-06-01 | Tuition — June | $1,240.00 |
| 2026-05-01 | Tuition — May | $1,240.00 |
The gap
Aggregators cover the banks. Everything else is still a manual download.
Bank feeds work because someone enumerated the banks. Your childcare portal, the HOA, the regional water utility, the marketplace that pays you out — nobody enumerated those, and nobody is going to. There are too many, each with too few customers.
So every month you sign in to each one, find the statement, and copy the rows out by hand. Hubdoc retiring Auto-Fetch made that worse for a lot of practices, and nothing replaced it for the long tail.
An institution list can never enumerate these sites. A workflow you taught your own browser doesn’t have to.
How it works
You do it once. Your browser does it from then on.
No site list to wait for, no integration to request. If you can reach the data signed in as yourself, it can be taught to reach it again.
- 01
Record it once
Do the thing you always do — sign in, pick the account, set the dates — and read back the steps in plain language. Passwords and one-time codes are never recorded: each becomes a pause that hands the keyboard back to you.
- 02
Replay it whenever
It drives your own tab the way you would, in a locked-down box that can only reach the site you taught it. When a portal is redesigned, it repairs the step that broke rather than leaving you a broken button.
- 03
Land it in your sheet
Each run adds what is new and corrects what changed, so running it twice does not give you everything twice. Columns you added yourself are left alone, and anything that would delete a row waits for your approval.
| A | B | C | D | |
|---|---|---|---|---|
| Date | Description | Amount | Categoryyours | |
| 2 | 2026-07-01new | Tuition — July | 1,240.00 | Childcare |
| 3 | 2026-06-03 | Late pickup fee | 35.00 | Childcare |
| 4 | 2026-06-01 | Tuition — June | 1,240.00 | Childcare |
| 5 | 2026-05-01 | Tuition — May | 1,240.00 | Childcare |
Which account, which months — anything that changes between runs is asked on your machine when the run starts rather than frozen into the recording. That is what makes the same recording work next month.
Privacy
We designed ourselves out of the data path.
Not a policy promise — an architecture. The extension has nowhere to send your records even if we wanted them, because the only place they go is a file you already own.
We never see your records
No transactions, balances, statements or logins. Rows go from the page to your sheet in your own session, with your own credentials. Our servers are not in the data path.
It all lives in your Drive
Your workflows and your spreadsheets are files in your own Drive — open them, edit them, move them, delete them. We keep no copy, so there is nothing of yours here to leak or to subpoena.
Nothing runs that you cannot read
Every step is written in plain language you can change or delete before it runs, and the run happens in a tab in front of you. What reaches us is templated telemetry with the values stripped out.
No evasion, ever
No disguising your browser, no routing through other people’s connections, no getting around the checks a site puts up. A product boundary rather than a roadmap gap: it is what separates automating your own accounts from scraping someone else’s site.
AI helps write and repair your workflows. It never runs them — once a workflow exists it replays on your machine, the same way every time, with no AI in the loop.
Coverage
Built for the long tail — on purpose.
We do not start with the major banks. They have the heaviest MFA, the strongest bot defence, and four products already serving them. The portals that bill you every month and export nothing are where a workflow you recorded yourself beats any institution list.
Childcare & education
Daycare portals, after-school programs, camps, tuition systems
Property & community
HOA portals, property managers, storage facilities
Utilities & municipal
Regional water, power and waste, city services, tolls
Small financial institutions
Regional credit unions, local banks, expat banks
Marketplaces & sellers
Seller centrals, payout portals, niche marketplaces
Professional services
Medical billing, veterinary, legal, insurance dashboards
Long-tail SaaS
Vendors that email nothing and only show invoices in-app
Curated recipes
Common portals get a reviewed template, reconstructed from aggregate structure — never copied from anyone’s workflow, and never a stranger’s code in your session.
Pricing
Priced by what you keep in sync, not by how often you check.
You already know how many accounts you pull from. That is the number you pay against — one per sheet you keep up to date, with runs unlimited on both plans.
Free
Start here$0
No card required, and no trial to run out.
- 2 sheets kept in sync
- Unlimited runs
- Unlimited workflows and one-off extractions
- AI authoring and repair included
- Curated recipes
Pro
$5.99/ month
For a monthly close that depends on this.
- 50 sheets kept in sync
- Unlimited runs
- Unlimited workflows and one-off extractions
- Generous AI allowance
- Priority repair when a portal changes
- Curated recipes
Upgrade inside the extension when you need a third sheet. It goes on the account you already signed in with.
Cancelling never stops a sheet you already set up — the limit governs new ones, so nothing you depend on goes quiet because a card expired.
Need more than 50 sheets? Talk to us.
Questions
The things people ask first.
Which browsers does it work in?
Chrome and other Chromium desktop browsers — Edge, Brave, Arc. It depends on Chrome extension features that Firefox and Safari do not offer, so those are out of scope, as is mobile: Chrome for Android does not support extensions at all.
Do you store my passwords?
No. Nothing you type is recorded, and any password or one-time-code field becomes a pause that hands control back to you mid-run. Other values you did record stay on your device and can be deleted — deleting one turns that step back into a pause for you, so the value is gone rather than hidden.
Is this a scraper?
It drives a browser you are already signed in to, at human pace, on accounts that are yours. Your login is never stored, nothing runs on our machines, and there is no getting around a site’s defences — no disguising your browser, no routing through someone else’s connection. If a site does not want to be automated, we do not have a workaround for you.
What about PDF invoices?
Downloaded documents are filed as-is for recordkeeping. We extract the tabular data already visible on the page — invoice totals, transaction rows, statement lines — and never parse a PDF into records.
Do the LLMs see my data?
They see how a page is laid out and the steps being written, not your records — and they never run anything. Once a workflow exists it replays the same way every time, on your machine, with no AI involved.
What happens when a portal changes?
The run stops rather than writing something wrong, and it proposes a fix for the step that broke. A failed run does not spend your allowance — keeping workflows working is what you are paying for, so charging for breakage would be backwards.
Stop downloading the same statement every month.
Teach it one portal and see what happens next month. Free to start, and everything it writes stays in your own Drive.